One of the most important limitations concerning analog computation is the lack of a unifying paradigm. Indeed, while several models of analog computation are known, few connections were established between them. Here we explore this topic and show equivalence between two models, Shannon’s General Purpose Analog Computer (GPAC) and Computable Analysis. In particular, we show that if we modify the notion of computability in the GPAC, and add some mild conditions on the model, then we get exactly the class of functions computable according to Computable Analysis. This talk describes ongoing work with Olivier Bournez and Emmanuel Hainry (done under the supervision of M. Campagnolo and J. Buescu).